Mikrotik: Ping disappears to the gateway via cable. What could be the reason?

Question

network hardware

There is Mikrotik RB3011UIAS-RM, firmware 6.47.10 stable, 2 WAN from two different providers, microt address 192.168.1.1, bridge 192.168.1.1/23. In all offices there is an unmanaged switch, which receives a cable data transmission channel (CPD) from the provider. The cAP ac points are connected to these unmanaged switches in the cabinets and through CAPSMAN they work in a seamless network. Microt on DHCP distributes addresses to all clients.
Microta ports:
eth1 - WAN from ISP #1
eht2 - WAN from ISP #2
eth5 - data link to ISP #1.
eth9 - directly connected computer 192.168.0.30

• computer 192.168.0.30 (connected directly to microt eht9) always pings gateway 192.168.1.1, printer 192.168.1.230 and point 192.168.0.253 always
  
• laptop 192.168.1.101 (connected via wifi to a point that is connected to a switch that is connected to a router through efficiency) pings both the gateway 192.168.1.1 and the printer 192.168.1.230 and the point 192.168.0.253 always
  

BUT periodically during the working day

• computer 192.168.0.40 (connected to the switch, which is connected to the router through the ETC) stops pinging the gateway 192.168.1.1, but at the same time it pings the printer 192.168.1.220 and the point 192.168.0.252, which are in another cabinet (this cabinet is also connected through the ETC)
  
• laptop 192.168.1.100 (connected to the switch, which is connected to the router through the ECM) stops pinging the gateway 192.168.1.1, but at the same time the printer 192.168.1.220 and the point 192.168.0.252, which are in another cabinet (this cabinet is also connected through the ECM)
  

And such a picture in all offices. In the middle of the day, computers in offices connected by cable stop pinging the gateway. Moreover, in one office, computer No. 1 pings the gateway, but as soon as it goes to sleep or distorts the cable to the ethernet connector, the ping to the gateway disappears right there. It turns out that some of the computers in the same office for some time with the Internet, and some of the computers / laptops (came out of sleep, or the laptop returned from lunch) without it. And in the end, the Internet disappears via cable in all offices that are connected by efficiency. At the same time, the computer connected directly to the mikrotik ping to the gateway has the Internet too.

After 19-20 in the evening (the main part of the BC employees leave), the ping itself appears in all offices without any reboots of the mikrotik or unmanaged switches.

The provider came to test, went crazy with this problem and said that it was a matter of Mikrotik or some kind of loop. He said that their efficiency is pure L2, and there are no problems with it. At the same time, the provider looked at the Mikrotik settings and said that everything seemed to be OK.

Today, in turn, in the middle of the day, we turned off / on the switches in all offices for 5 minutes (trying to find a loop or an enraged device) and pinged the gateway from another office - there was no ping.

It is clear that the network on unmanaged switches is bad, but there are such features of the BC that they do not allow to lay their own SCS, but force them to use the current one. Given all of the above, what are the possible solutions to the problem? There seem to be no loops, there are no errors in the microt logs, the provider says everything is in order with their efficiency ... Please advise where to dig.

Answer 1

Please advise where to dig

Dig towards the conflict of IP addresses (issued via DHCP).
For example, when 192.168.1.100 stops pinging the gateway, try pinging 192.168.1.100 from the gateway.
And from different other computers to ping 192.168.1.100.
And if such a ping passes, then disconnect (physically) 192.168.1.100 from the network and check the pings to it again from different computers.
Suddenly find that someone at this address answers, even when the computer is disconnected from the network.

Answer 2

In the scheme, a black hole with the inscription "Provader # 1", you cannot cope without the participation of third-party forces.

Answer 3

Moreover, in one office, computer No. 1 pings the gateway, but as soon as it goes to sleep or distorts the cable to the ethernet connector, the ping to the gateway disappears right there.

admin mac on bridge static?

Answer 4

How many devices are on the network, including WiFi clients? Providers may have a limit on the number of mac-addresses allowed in L2 VPN at the same time. Therefore, it is always better to keep some of your own routers at the ends of such channels, which will encapsulate the transmitted traffic. It would be good to clarify this point with the provider.
If there is a technical possibility, try using CAP points in this capacity (wrapping LAN traffic, for example, in EoIP), at least for the test, they have two wired interfaces, as far as I remember. By the way, CAPSMAN, depending on the settings, can also encapsulate WiFi client traffic.